Train Positive Dog!

Two weekends ago, I attended my sixth ClickerExpo.  My mind is still reeling with information and insights gained during the conference.  The picture above is from the pre-conference field trip to Guide Dogs for the Blind in Boring, Oregon (paired with Dull, Scotland).

The guide dog in training is being introduced to overhead barriers (barriers that the dog could go under, but the handler could not) for the first time.  This is one of the areas where clicker training has greatly facilitated the training of the guide dogs.  Using the clicker and food reinforcement, it is easy to mark the instant when the dog notices the barrier.  The bar of the barrier is moveable so that it is initially right in the dog’s face, and then is moved upwards, always striving to keep the dog successful in learning.

It was fascinating to watch this dog quickly and happily learn that he needed to stop at an overhead barrier.  Intelligent disobedience is another area where training was made easier and faster by using the clicker to mark (and then reinforce with food) when the dog made the right decision, even if the handler is cuing the dog to do something different.  Guide Dogs for the Blind uses clicker training extensively and works to maximize positive reinforcement in all aspects of training.

Two fascinating presentations at ClickerExpo were about what happens when reinforcement stops.  The first was a talk about resurgence by Jesus Rosales-Ruiz (University of North Texas).  Resurgence, the reappearance of behavior, occurs as follows –

“If present behavior is not capable of getting reinforcement, one reverts to older forms of response which were once effective”

Examples of resurgence are when the candy machine doesn’t dispense your treat and you start to pound on it or kick it; or when you suddenly stop giving click/treats in a training session and ignore your dog, and then he starts to offer uncued behaviors.  These extinction events are happening all the time to us and our dogs.  And the order in which behaviors reappear is very predictable – whether you are a dog or a human.  This predictable order of resurgence can be used during training.  Frustration can be avoided by training a cue that signals the end of a session and being aware of the effects of suddenly withholding reinforcement.

Michele Pouliot (Guide Dogs for the Blind) spoke about what happens when clicks are withheld.  Of course, there are, naturally, spaces between clicks and other reasons why you don’t click:

• you are extending the duration of a behavior
• you are selectively clicking only faster behaviors
• you are shaping behavior and waiting for the offered behavior

Withholding the click should give information to the dog, and prompt him to offer versions of the most recently clicked behavior enthusiastically.  It should inform the dog:  it wasn’t that, try something else.   If the dog is frustrated and disengaged, or offers a host of different behaviors, he is not ready for clicks to be withheld in that context.

As mentioned earlier, ending a training session will mean that reinforcement (click/treats) will end.  Although in clicker training we usually want to maintain a high rate of reinforcement; to avoid possible frustration, don’t end a training session with your dog at a high level of excitement. Think of the last few minutes as a cool-down period, especially if you have a dog who is easily excited and impulsive.  And train a clear end-of-session cue:

1. Use a unique visual cue such as crossing and uncrossing your arms
2. Use a unique verbal cue such as “All Done!”
3. Scatter a few treats on the ground for your dog to clean up
4. Leave the training area, perhaps taking your dog outside or to another room

It doesn’t mean that you can’t reinforce good behavior outside of planned training sessions, you should certainly do that.  Just be aware that intense training sessions may set your dog up for frustration if communication from you is not clear.  Just as you should warm up your dog when you start a session, you should cool down and signal the end.  What do you do to signal the end of a training session?